[Feature] Repair JSON from LLM in StructuredOutputParserAdvanced (#3723)

Repair JSON from LLM in StructuredOutputParserAdvanced

Co-authored-by: Henry <[email protected]>

Author: JJK801

packages/components/nodes/outputparsers/StructuredOutputParser/StructuredOutputParser.ts

@@ -3,6 +3,7 @@ import { BaseOutputParser } from '@langchain/core/output_parsers'
 import { StructuredOutputParser as LangchainStructuredOutputParser } from 'langchain/output_parsers'
 import { CATEGORY } from '../OutputParserHelpers'
 import { convertSchemaToZod, getBaseClasses, INode, INodeData, INodeParams } from '../../../src'
+import { jsonrepair } from 'jsonrepair'
 
 class StructuredOutputParser implements INode {
     label: string
@@ -74,6 +75,14 @@ class StructuredOutputParser implements INode {
             const zodSchema = z.object(convertSchemaToZod(jsonStructure)) as any
             const structuredOutputParser = LangchainStructuredOutputParser.fromZodSchema(zodSchema)
 
+            const baseParse = structuredOutputParser.parse
+
+            // Fix broken JSON from LLM
+            structuredOutputParser.parse = (text) => {
+                const jsonString = text.includes('```') ? text.trim().split(/```(?:json)?/)[1] : text.trim()
+                return baseParse.call(structuredOutputParser, jsonrepair(jsonString))
+            }
+
             // NOTE: When we change Flowise to return a json response, the following has to be changed to: JsonStructuredOutputParser
             Object.defineProperty(structuredOutputParser, 'autoFix', {
                 enumerable: true,

packages/components/nodes/outputparsers/StructuredOutputParserAdvanced/StructuredOutputParserAdvanced.ts

@@ -3,6 +3,7 @@ import { BaseOutputParser } from '@langchain/core/output_parsers'
 import { StructuredOutputParser as LangchainStructuredOutputParser } from 'langchain/output_parsers'
 import { CATEGORY } from '../OutputParserHelpers'
 import { z } from 'zod'
+import { jsonrepair } from 'jsonrepair'
 
 class AdvancedStructuredOutputParser implements INode {
     label: string
@@ -62,6 +63,14 @@ class AdvancedStructuredOutputParser implements INode {
         try {
             const structuredOutputParser = LangchainStructuredOutputParser.fromZodSchema(zodSchema)
 
+            const baseParse = structuredOutputParser.parse
+
+            // Fix broken JSON from LLM
+            structuredOutputParser.parse = (text) => {
+                const jsonString = text.includes('```') ? text.trim().split(/```(?:json)?/)[1] : text.trim()
+                return baseParse.call(structuredOutputParser, jsonrepair(jsonString))
+            }
+
             // NOTE: When we change Flowise to return a json response, the following has to be changed to: JsonStructuredOutputParser
             Object.defineProperty(structuredOutputParser, 'autoFix', {
                 enumerable: true,

packages/components/package.json

@@ -92,6 +92,7 @@
         "ioredis": "^5.3.2",
         "jsdom": "^22.1.0",
         "jsonpointer": "^5.0.1",
+        "jsonrepair": "^3.11.1",
         "langchain": "^0.3.5",
         "langfuse": "3.3.4",
         "langfuse-langchain": "^3.3.4",